FEBS Letters 589 (2015) 1450–1458

journal homepage: www.FEBSLetters.org

Induction of HSPA6 (HSP70B0) upon inhibition in cancer cell lines ⇑ Petric Kuballa a, , Anna-Lena Baumann a, Klaus Mayer a, Ute Bär b, Helmut Burtscher b, ⇑ Ulrich Brinkmann a, a Roche Pharma Research & Early Development, Large Molecule Research, Roche Innovation Center Penzberg, Nonnenwald 2, 82377 Penzberg, Germany b Roche Pharma Research & Early Development, Discovery Oncology, Roche Innovation Center Penzberg, Nonnenwald 2, 82377 Penzberg, Germany article info abstract

Article history: Genome-wide transcript profiling to elucidate responses to HSP90 inhibition revealed strong induc- Received 20 February 2015 tion of HSPA6 in MCF-7 cells treated with 17-AAG. Time- and dose dependent induction of HSPA6 Revised 10 April 2015 (confirmed by qPCR and Western Blots) occurred also upon treatment with Radicicol, another Accepted 27 April 2015 HSP90 inhibitor. HSPA6 was not detectable in untreated cells or cells treated with toxins that do Available online 7 May 2015 not inhibit HSP90, or upon applying oxidative stress. Thus, HSPA6 induction is not a general Edited by Lukas Huber response to cytotoxic insults. Modulation of HSPA6 levels by siRNA-mediated inhibition or recombi- nant expression did not influence 17-AAG mediated cell death. HSPA6 induction as a consequence of HSP90 inhibition occurs in various (but not all) cell lines and may be a more specific marker for Keywords: Cancer therapy HSP90 inhibition than induction of other proteins. Heat shock response Ó 2015 The Authors. Published by Elsevier B.V. on behalf of the Federation of European Biochemical Unfolded protein response Societies. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ Geldanamycin licenses/by-nc-nd/4.0/). Radicicol Oxidative stress

1. Introduction potency of Geldanamycin [9]. In complete agreement with that, a recent report demonstrates enhanced anti-cancer efficacy upon Geldanamycin is a small molecule that inhibits the function of dual targeting of HSP90 and HSP70 in some cancer subtypes [10]. HSP90 family proteins [1,2]. As HSP90 is overexpressed in many Therefore, a more detailed understanding of the interplay between tumors, Geldanamycin is a candidate for anti-cancer therapies members of the cellular HSP network in the context of [3], and efficacy of Geldanamycin derivatives is evaluated in HSP90-inhibition may improve anti-cancer therapies. numerous clinical trials [4–6]. Preliminary results suggest that The HSP70 family comprises at least 13 members [11], of which growth of HER2-positive breast cancers is particularly sensitive HSPA6 (HSP70B‘) appears evolutionary unique as it is not towards HSP90 inhibition by the Geldanamycin derivative conserved in rodents [12,13]. HSPA6 expression is not detectable 17-AAG (Tanespimycin) [7]. This observation might be explained in most cells under normal conditions. It becomes induced upon by the dependency on HER2 expression of this cancer subset as severe stress conditions and might mediate cytoprotective func- HER2 is a ‘‘client’’ protein of HSP90 [8]. HSP90 is a master regulator tions in a cell-type and context-dependent manner [11,14,15]. of HER2 protein stability, and inhibition of HSP90 function results The potential influence of drug-mediated HSP90 inhibition on in degradation of HER2 [8]. It appears, however, that some mem- expression of HSPA6 has not been analyzed so far. This work bers of the Heat shock 70 protein family become induced in describes the specific induction of HSPA6 upon treatment of cancer Geldanamycin treated cells which interferes with the therapeutic cells with HSP90 inhibitors.

2. Results Author contributions: P.K., experimental strategy, designed and performed experi- ments, wrote paper; U.Br., project concept & design, experimental strategy, wrote 2.1. Geldanamycin induces the expression of HSPA6 in MCF-7 breast paper; AL.B., U.Ba, K.M., H.B., designed and performed experiments, contributed to cancer cells paper. ⇑ Corresponding authors. E-mail addresses: [email protected] (P. Kuballa), ulrich.brinkmann@ A genome-wide analysis of transcriptional responses in MCF-7 roche.com (U. Brinkmann). breast cancer cells was performed upon exposure to respective http://dx.doi.org/10.1016/j.febslet.2015.04.053 0014-5793/Ó 2015 The Authors. Published by Elsevier B.V. on behalf of the Federation of European Biochemical Societies. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458 1451

IC50 concentrations of the HSP90 inhibitor 17-AAG, in comparison if GA-induced HSPA6 transcription translates to HSPA6 protein to other cytotoxic compounds with cellular targets other than production. We separated 17-AAG treated and untreated MCF7 cell HSP90: these included alpha-amanitin (ama, targets RNA-Pol), extracts on SDS/PAGE followed by immunoblotting with a saporin or cycloheximide (sap, CHX, both target the ribosome). HSPA6-specific antibody. HSPA6 protein is not detectable in We were able to recapture previously described untreated, but in 17-AAG-treated cells (Fig. 2A). Similar to our Geldanamycin-mediated effects, such as the induction of various observations for HSPA6 mRNA levels, HSPA6 protein levels reach HSP70 and HSP40 family members [16] (Table 1), thus validating a plateau at around 4 h following 17-AAG treatment (Fig. 2A) and our experimental work-flow. Surprisingly, we also observed that induction is concentration-dependent up to 500 nM 17-AAG HSPA6, a poorly characterized member of the HSP70 family (Fig. 2B). In contrast to other HSP70 proteins (the became induced to higher levels (more than 3000-fold) than any HSP70-antibody used in this study detects HSP72, HSPA1L as well other (up to 50-fold), even though induction of HSPA6 expres- as HSPA8), which remain significantly enriched even 24 h after sion upon Geldanamycin treatment has not been described before. 17-AAG treatment, HSPA6 protein levels decline rapidly between HSPA7 is very closely related to HSPA6 (the putative HSPA7 protein the 8 h and 24 h time points (Fig. 2A). is 94% identical to HSPA6), but has been characterized as a pseudo- Taken together, we conclude that exposure of MCF-7 cells to gene [13,17]. Therefore, the mRNA hybridization signals could 17-AAG leads to a transient induction of HSPA6 protein expression. have been derived from HSPA6 and/or HSPA7. To differentiate between these two transcripts, we performed qPCR with three 2.2. HSPA6 expression is specifically induced by HSP90 inhibitors different primer sets at different time points following 17-AAG treatment. Each of these primer sets amplified regions of A comparison of the genome wide mRNA profiles of MCF7 cells HSPA6/HSPA7 with at least one nucleotide mismatch between treated with various toxins reveals that HSPA6 signals become HSPA6 and HSPA7 sequences. We were subsequently able to dis- induced only by 17-AAG exposure, but not by toxins with targets tinguish between HSPA6/HSPA7 transcripts by sequencing tran- other than HSP90 (Table 1). This indicates that HSPA6 induction script fragments derived from qPCRs at 4 h time point (=maximal is not a general response towards cytotoxic insults, but instead induction). This revealed that all amplified fragments exclusively either a specific consequence of treating cells with the substance contained HSPA6 sequences, indicating exclusive (or at least pre- 17-AAG, or triggered by inhibition of its molecular target HSP90. dominant) induction of HSPA6 and not HSPA7 (Fig. 1). These In order to determine if induction of HSPA6 is a direct conse- qPCR analyses also revealed that induction of HSPA6 transcripts quence of HSP90-inhibition, we examined the effects of Radicicol, is time-dependent. It peaks around 4 h and declines to low levels another HSP90-inhibitor, on HSPA6 expression in MCF-7 cells. between 4 and 8 h after 17-AAG treatment (Fig. 1). Further, Similar to 17-AAG, Radicicol induced HSPA6 mRNA (Fig. 3)ina HSPA6 mRNA induction is dose-dependent and saturation is concentration-dependent manner. Maximum levels of induced observed with concentration of 17-AAG between 125 and HSPA6 transcripts appear somewhat lower in Radicicol- compared 250 nM (Fig. 1C). Next, we performed Western Blot analyses to test to 17-AAG-treated cells. This may be caused by the previously

Table 1 Induction of HSPA6 in 17-AAG treated MCF-7 cells. Cells were exposed for the indicated time periods to IC50 concentrations of toxins that either inhibit protein synthesis (Sap = Saporin, CHX = Cycloheximide), protein (re-)folding (17-AAG = 17-N-allylamino-17-demethoxygeldanamycin), or transcription (Ama = alpha-amanitin). Listed are log change values, i.e. fold changes are calculated according to the formula (Euler Constant)value. Changes were observed with at least 2 probes for HSPA6/HSPA7, DNAJ and HSP90. HSP70A, HSP70B, and HSP70L were detected with one specific probe for each mRNA and one probe that detects HSP70A as well as HSP70B. Increased expression is indicated green (positive values), decreased expression in red (negative values), minor or no changes in yellow. 1452 P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458

A C

B

Fig. 1. Time- and dose-dependent induction of HSPA6, but not HSPA7, mRNA by 17-AAG. (A) MCF-7 cells were treated with 500 nM 17-AAG and mRNA extracted at different time points, as indicated, after treatment. Following reverse transcription, qPCR assays were performed using 3 different primer sets for HSPA6/HSPA7. Amplification of the housekeeping gene GAPDH served as an internal control and reference. Values for HSPA6/HSPA7 represent fold-changes compared to expression in untreated cells. Statistical analyses (one-way ANOVA, unpaired 2-tailed t-test) reveal significant time-dependent differences of HSPA6 induction vs control with for all time points vs each other and primer sets (P < 0.01 for all steps and primers except primer set 2 for 2–4 h P < 0.05). Induction levels decrease significantly after 4 h (P < 0.01 4 h vs 8 h for all primer sets). (B) PCR products derived from qPCR assays performed on samples from cells treated for 4 h with 17-AAG treatment (see (A)) were separated on a 2% Agarosegel, extracted, column-purified and sequenced. Each primer set generated a single PCR-product of the expected size (224 bp for primer set1, 94 bp for primer set2, and 115 bp for primer set3). Sequencing confirmed the specific amplification of HSPA6 as all sequences aligned to 100% with the reference sequence of HSPA6 and positions with mismatches (marked ‘x’) for HSPA6 and HSPA7 did not amplify any signals, which could be derived from HSPA7 transcripts. The cDNA sources for HSPA6 were NCBI refseq NM_002155 and Genbank_AF093759 for HSPA7. The numbering of the HSPA6 and HSPA7 sequences relate to the nucleotide numbers in these sequences. (C) MCF-7 cells were treated for 4 h with different concentrations of 17-AAG, as indicated. QPCR assays were performed with HSPA6 primer set3 and GAPDH to determine fold-changes of HSPA6 transcripts compared to untreated cells. Statistical analyses (one-way ANOVA, unpaired 2-tailed t-test) reveal significant dose-dependent increases of HSPA6 induction vs control (DMSO) with P < 0.03 for 62.5 nM, P < 0.02 for 62–125 nM. Induction reaches saturation after 125 nM (therefore induction levels do not increase further in a significant manner).

AB

Fig. 2. Time- and dose-dependent induction of HSPA6 protein by 17-AAG. (A) MCF-7 cells were treated with 500 nM 17-AAG and cells lysed at different time points,as indicated, after treatment. Samples were separated on SDS–PAGE, proteins transferred to a PVDF membrane and HSPA6, HSP70 and Actin proteins detected with appropriate antibodies. (B) MCF-7 cells were treated for 8 h with different concentrations of 17-AAG, as indicated, cells lysed and processed as described in (A). described mRNA destabilization effects mediated by Radicicol represents a specialized, differentially regulated arm of the HSP70 derivatives [18]. family or if HSPA6 is commonly co-induced along with other These results indicate that the specific trigger of HSPA6 induc- HSP70 proteins. 17-AAG has been reported to cause ER stress and tion is the inhibition of HSP90. to induce of Unfolded Protein Response (UPR) pathway, including the HSP70 family member GRP78/BiP (also known as 2.3. HSPA6 is not induced by Thapsigargin- or BrefeldinA-mediated HSPA5 or HSP70-5) [19,20]. Further, GRP78/BiP is induced by other induction of the unfolded protein response ER stress/UPR inducing compounds with modes of action other than HSP90 inhibition, e.g. Thapsigargin [21,22] and BrefeldinA [23,24]. Cytotoxins, other than 17-AAG, used in our genome-wide tran- Thus, we were wondering if treatment of cells with Thapsigargin scriptional analysis did not significantly induce Heat shock 70 fam- or BrefeldinA may also result in induction of HSPA6. Interestingly, ily proteins (Table 1). Therefore, our data do not indicate if HSPA6 only 17-AAG and Radicicol, but not Thapsigargin or BrefeldinA P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458 1453

Fig. 3. Dose dependent induction of HSPA6 mRNA by Radicicol. MCF-7 cells were treated for 4 h with different concentrations of Radicicol, as indicated, or 1 lM of 17-AAG. Subsequently, qPCR assays were performed as described in Fig. 1. Shown are fold-changes for HSPA6 compared to untreated cells. Statistical analyses (one-way ANOVA, unpaired 2-tailed t-test) reveal significant dose-dependent increases of HSPA6 induction vs lowest AAG concentration (12.3 nM) with P < 0.02 for 37 nM, P < 0.03 for 37– 111 nM, P = 0.054 for 333 nM and higher. Induction reaches saturation after 111 nM (therefore induction levels do not increase further in a significant manner).

that the trigger of HSPA6 induction is specifically and directly linked to HSP90 inhibition, and not a consequence of UPR induction. Also, expression of the HSP70 family proteins HSPA6 and GRP78/BiP does not seem to be co-regulated. However, increased expression of ‘‘HSP70’’ (‘‘HSP70’’ = simultaneous detection of HSP72, HSPA1L and HSPA8) is observed following treatment with 17-AAG and Radicicol, but not Thapsigargin or BrefeldinA. Thus, induction of HSPA6 expression correlates with upregulation of one or more other HSP70 family proteins (Fig. 4 and Table 1). Like other HSP70 family members, HSPA6 becomes induced upon applying heat-shock (Fig. 9 and Suppl. Fig. 5), most likely as consequence of HSF-dependent transcription.

Fig. 4. HSPA6 induction is specifically triggered by HSP90 inhibition. MCF-7 cells 2.4. Induction of HSPA6 expression following HSP90 inhibition in were treated for 6 h with approximately 4xIC50 concentrations of 17-AAG (1 lM), different cancer cell lines Radicicol (1 lM), Thapsigargin (30 nM) or BrefeldinA (0.1 lg/ml). Following cell lysis, samples were separated on SDS–PAGE, proteins transferred to a PVDF Is HSPA6 induction upon HSP90 inhibition a specific feature membrane and HSPA6, GRP78/Bip, HSP70 and Actin detected with appropriate antibodies. of MCF-7 (breast cancer) cells, or does it represent a more gen- eral response that can be observed in other cells? To address this question, we treated 12 different cancer cell lines with their can induce HSPA6 protein expression (Fig. 4). In contrast, the UPR respective IC50 concentrations of 17-AAG (Fig. 5, Table 2 and marker protein GRP78/BiP is strongly upregulated following treat- Suppl. Data S1) and determined HSPA6 levels by Western ment with any of the 4 compounds tested (Fig. 4). This indicates Blotting. Of the 7 breast cancer cell lines that we analyzed, 4

A

- + - + - + - + - + - + - + - + - + - + 75 HSPA6 50 Acn

BCSK-BR3 MCF-7 Raji MCF-7 - + - + 75 75 HSPA6 HSPA6

50 50 Acn Acn

Fig. 5. Induction of HSPA6 by 17-AAG in various cancer cell lines. (A–C) Cancer cell lines were treated for 6 h with their respective IC50 concentration of 17-AAG (see Fig. S1 and Table 2). HSPA6 and Actin protein levels were determined by Western Blot analysis. (B) In addition, 5xIC50 concentrations of 17-AAG have been applied to SK-BR3 cells to determine if lack of HSPA6 induction is cell line or concentration dependent. (C) Raji and MCF-7 cells have been treated with 20 lM 17-AAG (=estimated IC50 for Raji) for 6 h to exclude detrimental effects on HSPA6 induction with very high concentrations of 17-AAG. 1454 P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458

Fig. 6. ATP levels in MCF-7 cells following RNA Interference to blank HSPA6 induction in response to 17-AAG. MCF-7 cells were transfected with non-targeting (control) or HSPA6-specific (#1–3) siRNA duplexes. (Upper panel), 4 h after transfection, cells were treated with 1 lM 17-AAG. 4 h after 17-AAG treatment cells were lysed, samples separated on SDS–PAGE, and Actin and HSPA6 proteins detected by Western Blotting. (Lower panel), 4 h after transfection, cells were treated with increasing concentrations of 17-AAG, as indicated. 72 h after 17-AAG treatment, Cell Titer Glo assays were performed.

Fig. 7. BrdU incorporation in MCF-7 cells following RNA Interference to blank HSPA6 induction in response to 17-AAG. MCF-7 cells were transfected with control siRNA (‘‘C’’) or HSPA6-siRNA#3 (‘‘KD’’). (Upper panel), 4 h after transfection, cells were treated with increasing concentrations of 17-AAG, as indicated. 4 h after 17-AAG treatment cells were lysed, samples separated on SDS–PAGE, and Actin and HSPA6 proteins detected by Western Blotting. (Lower panel), 4 h after transfection, cells were treated with increasing concentrations of 17-AAG, as indicated. 24 h after 17-AAG treatment, BrdU incorporation assays were performed.

(MCF-7, KPL4, BT-474, MDA-MB-361) demonstrated a strong MDA-MB-231, SK-BR3 and Raji cells which are rather unrespon- induction of HSPA6 protein. The other three breast cancer cell sive to 17-AAG and require high concentrations of 17-AAG to lines (MDA-MB-134-VI, MDA-MB-231, SK-BR3) showed no induce cell death do not induce HSPA6. Three of six cell lines induction of HSPA6 protein upon HSP90 inhibition (Fig. 5). Five with medium sensitivity also lack HSPA6 induction. In contrast, additional cell lines with tissue origins other than breast were the three cell lines that are most sensitive to 17-AAG also examined. Among these, 17-AAG exposure does result in (MDA-MB-361, Colo-205 and BT-474) are HSPA6 inducers strong HSPA6 induction in A-431 (epidermoid carcinoma) and (Table 2 and Suppl. Fig. 2). However, lack of HSPA6 induction Colo-205 (colon cancer) cells, but not in Hep-G2 (liver cancer), may not reflect a general defect in mediating effects of HSP90 Molm-13 (Acute Myeloid Leukemia) and Raji (Burkitt’s inhibition as induction of HSP70, but not HSPA6, is observed in Lymphoma) cells (Fig. 5). Interestingly, the 3 cell lines Raji cells (Suppl. Fig. 4). P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458 1455

A

BC

Fig. 8. ATP levels and BrdU incorporation in MCF-7 cells following HSPA6 overexpression and 17-AAG exposure. MCF-7 cells were transfected with pCMV6-entry (‘‘empty vector’’) or pCMV6-HSPA6-untagged (‘‘HSPA6 vector’’) plasmids. 4 h after transfection medium was replaced. Another 20 h later, 17-AAG was added to increasing final concentrations, as indicated. (A), 4 h after 17-AAG exposure, cells were lysed, samples separated on SDS–PAGE, and Actin and HSPA6 proteins detected by Western Blotting. (B), 72 h after 17-AAG treatment, Cell Titer Glo assays were performed. (C), 24 h after 17-AAG treatment, BrdU incorporation assays were performed.

Fig. 9. MDA-MB-231 or MCF-7 cells were treated for 6 h respective 4xIC50 concentration of various compounds, as indicated. HSPA6 and Actin protein levels were determined by Western Blot analysis.

2.5. Expression levels of HSPA6 do not affect cellular sensitivity to 17- Table 2 AAG IC50 concentrations of 17-AAG and HSPA6-responsiveness for all cell lines used in ⁄ this study. ( ) Fold-induction of HSPA6 protein is estimated based on serial dilution of The observation that the three cell lines that were least sensi- HSPA6 signals in 17-AAG-treated Colo-205 cells (Suppl. Fig. 3). tive to 17-AAG lacked HSPA6 induction and three cell lines that Cell line Cancer IC50 (nM) HSPA6 were most sensitive to 17-AAG showed strong HSPA6 induction, ⁄ induction ( ) raises the question if HSPA6 levels directly affect the cytotoxicity MDA-MB-361 Breast 55 +(>8) of 17-AAG. To this end, we transfected a scrambled siRNA or 3 dif- BT-474 Breast 100 +(>16) ferent HSPA6-specific siRNAs into MCF-7 cells. All HSPA6-specific Colo-205 Colon 125 +(>100) KPL-4 Breast 250 +(>16) siRNA sequences efficiently blocked induction of HSPA6 upon MCF-7 Breast 250 +(>32) treatment with 1 lM 17-AAG (Fig. 6, upper panel). Ablation with MDA-MB-134-VI Breast 250 – HSPA6-siRNA#3 demonstrated an efficient suppression of HSPA6 Molm-13 Acute Myeloid Leukemia 300 – induction even at concentrations of 17-AAG up to 4 lM(Fig. 7). A-431 Epidermoid carcinoma 375 +(>4) Upon transfection of two siRNAs (HSPA6-siRNA#2 and 3) a modest Hep-G2 Liver 375 – MDA-MB-231 Breast 2000 – ‘non-specific’ reduction of cell metabolism (ATP content, Fig. 6) SK-BR3 Breast 15000 – was observed in the absence of 17-AAG. However, HSPA6-siRNAs Raji Burkitt’s lymphoma 20000 – did not affect cell proliferation rates (BrdU incorporation, Fig. 7). 1456 P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458

Furthermore, exposure of HSPA6-siRNA treated cells to increas- inhibition by a (possibly compensatory) upregulation of other heat ing concentrations of 17-AAG revealed no difference in cellular shock proteins, particularly members of the HSP70 family [9,10]. sensitivity between cells that induced HSPA6 and cells that did Thus, a detailed knowledge of the interplay between members of not induce HSPA6 (Figs. 6 and 7). Next, we were wondering if the cellular HSP network in the context of HSP90-inhibition might HSPA6 function might be relevant under conditions of very severe help in the design of strategies that avoid the activation of escape stress. Therefore, we repeated our knock down experiments and mechanisms by cancer cells. We report that in various cancer cell challenged cells by heat shock at 42 °C with and without additional lines, expression of HSPA6 protein is induced following administra- 17-AAG treatment. Heat shock and 17-AAG treatment alone or in tion of 17-AAG. This effect appears to be a specific consequence of combination resulted in induction of HSPA6 protein, which could HSP90 inhibition, as only HSP90 inhibitors (17-AAG and Radicicol), be virtually blanked by our HSPA6-siRNA (Suppl. Fig. 5, upper but not other cytotoxic compounds with different mode of actions, panel). Surprisingly, although HSPA6 induction seems to be very result in a similar induction of HSPA6. In MCF-7 cells and several strong upon heat stress in combination with 17-AAG challenge other cell lines, HSPA6 induction on mRNA and protein level is a (Suppl. Fig. 5, upper panel), ablation of HSPA6 does not affect cel- time- and concentration-dependent consequence of treatment lular sensitivity to both insults at any concentration of 17-AAG with 17-AAG or Radicicol. Some other cell lines, however, do not (Suppl. Fig. 5, lower panel). Taken together, ablation of HSPA6 elicit a ‘HSPA6 response’, suggesting that cell line dependent fac- induction did not protect against 17-AAG or 17-AAG plus heat tors may be implicated in the transcriptional control of HSPA6 shock mediated decline of cell proliferation or viability rates in downstream of HSP90. The molecular pathways of HSPA6 tran- MCF-7 cells. scriptional control are only beginning to be elucidated and control In addition, we also analyzed the effect of ectopic expression of of HSPA6 transcription is still poorly understood. A recent report HSPA6 on 17-AAG sensitivity in MCF-7 cells. Transfections condi- suggests the existence of positive as well as negative regulatory tions were optimized to generate steady-state levels of overex- elements in the HSPA6 promoter [25]. It is possible that one or pressed HSPA6 comparable to 17-AAG-induced levels of more HSP90 client protein(s) might be involved in the endogenous HSPA6 (Fig. 8, see Section 4 for transfection details). transcriptional regulation of HSPA6. Our data point towards a tran- Under the conditions used, ectopic expression of HSPA6 did not scriptional control of HSPA6 following HSP90 inhibition that is alter cell proliferation and viability in the absence or presence of different from the induction of HSPA6 in response to the protea- 17-AAG (Fig. 8). Further, MDA-MB-231 cells, which do not induce some inhibitor MG132 or heat stress. While heat shock induction endogenous HSPA6 in response to 17-AAG, do not show an altered of HSPA6 is mediated by the heat shock element (HSE), induction sensitivity towards 17-AAG when modified to stably overexpress of HSPA6 in response to MG132 is not defined on a molecular level. HSPA6 (Suppl. Fig. 6). Taken together, HSPA6 induction correlates HSE mediated transcription cannot fully explain HSPA6 regulation, to some extent with 17-AAG responsiveness, but does not seem because some cell lines are fully competent to induce HSPA6 dur- to represent a survival factor or component that ameliorates ing heat stress, but do not induce HSPA6 upon HSP90 inactivation. 17-AAG toxicity. Recently a functional AP-1 site within the HSPA6 promoter region was described and MG132 is known to induce transcriptional 2.6. MG132 and heat stress induce HSPA6 in cells that do not induce activity from AP-1 promoter sequences [25]. Thus, HSP90 inhibi- HSPA6 in response to 17-AAG tion, proteasome inhibition and heat shock might mediate HSPA6 induction by distinct mechanisms, of which only signaling down- HSPA6 induction following heat stress or treatment of cells with stream of HSP90 appears to be highly cell type specific. It would the proteasome inhibitor MG132 have been described before. As be of interest to determine if the postulated promoter elements these triggers do not act via HSP90 inhibition, we were wondering C/EBP or ZFX may be involved in HSPA6 induction downstream if these stimuli are effective for HSPA6 induction in cells that do of HSP90 [25]. not induce HSPA6 in response to 17-AAG. MDA-MB-231 cells fail It has been suggested that HSPA6 can form a complex with to induce HSPA6 in response to 17-AAG, but show a prominent HSP90, but no function has been assigned to this putative induction of HSPA6 following treatment with 4xIC50 concentration HSPA6/HSP90 complex. Interestingly, formation of HSPA6/HSP90 of MG132 (Suppl. Fig. 7) or exposure of cells to 42 °C for 1 h or 2 h complexes is Geldanamycin-dependent [26]. Given the absence (Fig. 9). In contrast, MCF-7 cells induce HSPA6 protein in response of HSPA6 expression in untreated cells, it is tempting to speculate to MG132, heat shock and 17-AAG (Fig. 9). This may suggest, that that Geldanamycin-dependency of this complex formation is sim- transcriptional upregulation of HSPA6 in response to 17-AAG is not ply reflected by Geldanamycin-dependent expression of HSPA6. solely dependent on the Heat Shock Element (HSE) within the MG132, a proteasome inhibitor representing another class of HSPA6 promoter region. In addition, we tested additional cytotoxic anti-cancer drugs also induces expression of HSPA6 in human cells compounds, that act in a HSP90-independent manner. No [27,28]. In this example, HSPA6 induction was suggested to be a induction of HSPA6 could be observed following exposure of cellular ‘pro-survival’ response, because increased sensitivity MCF-7 cells to respective 4xIC50 concentrations (Suppl. Fig. 7)of towards MG132-induced cell death has been reported for human the autophagosome-lysosome-fusion inhibitor BafilomycinA1, the cell lines lacking a ‘HSPA6 response’ [27]. In contrast, our data indi- lysosome inhibitor Hydroxychloroquine, the oxidative stress cate that cell lines with high tolerance for 17-AAG (SK-BR3, Raji inducer Hydrogenperoxide or the protein kinase inhibitor and MDA-MB-231) do not induce HSPA6 expression (Fig. 5). Staurosporine (Fig. 9). This further supports the notion that Furthermore, ablation of HSPA6 induction in 17-AAG sensitive HSPA6 induction is not a general response to cytotoxic insults, MCF-7 cells did not alter sensitivity towards 17-AAG-mediated cell but rather controlled in a very defined, yet complex, manner. death (Figs. 6 and 7). A similar finding has been reported for heat shock induced cell death, where neither knock down nor overex- pression of HSPA6 did affect cell viability [11]. Thus, cell type 3. Discussion specific and context dependent factors might contribute to the function and localization of HSPA6. Interestingly, localization to Geldanamycin derivatives, such as 17-AAG, are drug candidates the extracellular milieu [29], cell surface [28], centrioles [30] and for anticancer therapy [4–6]. The target of Geldanamycin, HSP90, is the nucleus [31], has been described for HSPA6. overexpressed in many cancer tissues studied to date [3]. However, Molecular markers to track HSP90 inhibition as well as more cancer cells seem to be able to escape cytotoxic effects of HSP90 detailed information about the consequences of HSP90 inhibition P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458 1457 are desired for basic research as well as in the clinic. HSPA6 mRNA (Light Cycler RNA Master SYBR Green I kit), and specific amplifica- and protein are virtually undetectable in untreated cells, but highly tion primers. QPCR was performed on a Lightcyler (Roche) applying induced in 17-AAG treated cells (up to 7000-fold induction of 3 different primer pairs for HSPA6 [(HSPA6-Primer Set1, forward: HSPA6 mRNA in MCF-7 cells, Fig. 1C). In contrast to general induc- CAAGGTGCGCGTATGCTAC; reverse: GCTCATTGATGATCCGCA tion of HSP70 proteins (with background levels and general ACAC), (HSPA6-Primer Set2, forward: CATCGCCTATGGGCTG GAC; UPR-induction (Fig. 2), determination of HSPA6 levels (mRNA or reverse: GGAGAGAACCGACACATCGAA), (HSPA6-Primer Set3, for- protein) can represent a very clean and highly sensitive readout ward: GATGTGTCGGTTCTCTCCATTG, reverse: CTTCCATGAAGT for successful HSP90 inhibition. Especially at early time points GGTTCACGA)] and the primers GAPDHfor (GAAGGTGAAGGT (2–8 h) following 17-AAG administration, fold-change is much CGGAGTCA) and GAPDHrev (GAAGATGGTGATGGGATTTCCA) to higher for HSPA6 as compared to other 17-AAG-responsive amplify the GAPDH housekeeping gene as normalization standard. genes/proteins, e.g. other HSP70 proteins (Table 1 and Fig. 2A). In subsequent experiments, HSPA6-Primer Set3 was used to quan- A restriction in applying HSPA6 induction as general marker for tify HSPA6 transcripts. HSP90 inhibition is our observation that not all cell lines or cell types induce HSPA6 following 17-AAG exposure. Thus, an initial 4.4. Western Blotting, antibodies and recombinant proteins evaluation step of HSPA6-responsiveness in a given cell type/cell line is required before applying HSPA6 as sensitive marker for For protein detection by Western Blotting, 600000 cells were HSP90 inhibition. plated on 6-well plates. The day after, test compounds were added to cells. At different time points, as indicated, cells were lysed in RIPA buffer (Sigma) supplemented with complete protease inhibi- 4. Materials and methods tor cocktail (Roche). Cell lysates were normalized for total protein amounts (BCA protein assay kit, Pierce) and resolved by SDS/PAGE 4.1. Cell culture (4.12% NuPAGE SDS–PAGE, Life Technologies). Following antibodies were used for immunodetection: MCF-7 [ATCC, HTB-22], KPL-4 [Kawasaki Medical School], Anti-HSPA6 (OriGene Technologies, mouse monoclonal, clone Hep-G2 [ATCC, HB-8065], Molm-13 [DSMZ, ACC-554], BT-474 1D12), Anti-Beta-Actin (Sigma, mouse monoclonal, clone AC-74), [ATCC, HTB-20], A-431 [ATCC, CRL-1555], Colo-205 [ATCC, HSP70 (Cell Signaling), Anti-BiP (Cell Signaling, rabbit mAb, clone CCL-222], MDA-MB-231 [ATCC, HTB-26], MDA-MB-361 [ATCC, C50B12), and HRP-coupled anti-mouse or anti-rabbit secondary HTB-27], MDA-MB-134-VI [ATCC, HTB-23], Raji [ATCC, CCL-86] antibody (Dako, goat polyclonal). Recombinant HSPA1A and and SK-BR3 [ATCC, HTB-30] cells were cultured at 37 °Cin5% HSPA8 used to determine specificity of HSPA6 and HSP70 antibod- CO atmosphere with 80% humidity. Following base media were 2 ies were purchased from origene (Catalog number TP10001 and used: RPMI (MCF-7, Colo-205, BT-474, Molm-13, Raji), DMEM TP302209, respectively). For Western Blot analysis 500 ng of (Hep-G2, A-431, KPL-4), L-15 (MDA-MB-134-VI, MDA-MB-231, recombinant protein was loaded per lane. The specificity of the MDA-MB-361) and McCoy’s 5a (SK-BR3). Media contained HSPA6 antibody was proven by several lines of evidence: knock- NaHCO at concentrations of 0.7 g/l (L-15 media), 2.0 g/l (RPMI 3 down experiments applying HSPA6 mRNA specific reagents (gene media) or 2.2 g/l (DMEM media). Further, all cell culture media specific sequences) prevents generation of antibody-signals in were supplemented with 10% fetal bovine serum and 2 mM otherwise responsive cells, ‘17AAG unresponsive’ cell lines can Glutamine. For culture of Molm-13 cells additional fetal bovine induce other HSP70 proteins upon 17AAG treatment but are still serum was added (final concentration = 20%) and culture media negative for HSPA6 (crossreactivity of the HSPA6 antibody with for BT-474 cells was supplemented with human Insulin (final other HSPs would generate signals under such conditions), and concentration = 10 lg/ml). Western Blots with HSP70 proteins HSPA1A (Hsp70-1) and HSPA8 (Hsc70) side by side with HSPA6 induced cell extracts 4.2. Affymetrix revealed no reactivity with both proteins. Transcriptional profiling was performed by chip-based (Affymetrix chip) hybridization analyses of mRNA isolated from 4.5. RNAi and recombinant expression MCF-7 cells as previously described [32]. In brief, cells were seeded and grown to subconfluency in 6-well plates and exposed to For RNA Interference the following stealth siRNAs from 17-AAG or other toxins, each at IC50 concentration. At different LifeTechnolgies were used: scrambled control siRNA (siRNA time points thereafter, RNAs were prepared from these cells and Negative Control Med GC Duplex #2, catalog number subjected to Affymetrix chip hybridization experiments according 12935-112), HSPA6-siRNA#1 (HSS105080), HSPA6-siRNA#2 to the manufacturer’s protocol. RNA from untreated MCF-7 cells (HSS179391) and HSPA6-siRNA#3 (HSS179392). MCF-7 cells were served as reference. All analyses were performed in triplicates, transfected with 30 pmol (60000 cells, 24 well format) or 5 pmol and passed quality controls for mRNA/cDNA preparation. (10000 cells, 96 well format) of respective siRNA duplexes with Lipofectamine 2000 according to manufacturer’s manual. 4 h after transfection cells were exposed to different concentrations of 4.3. RT-PCR 17-AAG, as indicated. For overexpression experiments plasmids encoding for no protein, herein referred to as ‘‘empty vector’’ For quantitative real time analysis of RNA transcripts, MCF-7 (pCMV6-entry, origene, SKU RC207795) or HSPA6 without any tags cells were seeded at 180000 cells per well on a 12-well plate. (HSPA6 (untagged), origene, SKU SC118784) were used. 400 ng The day after, test compounds were added to cells. At different (60000 cells, 24well format) or 66,7 ng (10000 cells, 96 well for- time points, as indicated, cells were lysed and RNA extracted mat) total DNA were transfected with Lipofectamine 2000 accord- applying the RNeasy Mini kit (Qiagen). Following removal of geno- ing to manufacturer´s manual. DNA mixes for HSPA6 mic DNA (TURBO DNA-free kit, Ambion), RNA was reverse tran- overexpression contained 1/8 of HSPA6 vector and 7/8 of empty scribed using a Transcriptor First Strand cDNA Synthesis Kit vector. 4 h after transfection medium was replaced and another (Roche). CDNA samples were mixed with SYBR Green Master Mix 20 h later 17-AAG was added. 1458 P. Kuballa et al. / FEBS Letters 589 (2015) 1450–1458

4.6. Quantification of cell numbers and cell proliferation -like activities. Biochem. J. 435, 127–142, http://dx.doi.org/10.1042/ BJ20101247. [12] Leung, T.K., Rajendran, M.Y., Monfries, C., Hall, C. and Lim, L. (1990) The human Cell numbers were detected based on total ATP levels (Cell Titer heat-shock protein family. Expression of a novel heat-inducible HSP70 Glo assay, Promega) and BrdU incorporation was used to quantify (HSP70B0) and isolation of its cDNA and genomic DNA. Biochem. J. 267, cell proliferation rates (BrdU, ELISA etc. kit, Roche). Cell Titer Glo 125–132. [13] Parsian, A.J. et al. (2000) The human Hsp70B gene at the HSPA7 locus of and BrdU incorporation assays were performed 72 h and 24 h, is transcribed but non-functional. Biochim. Biophys. Acta 1494, respectively, after exposure of cells to 17-AAG according to manu- 201–205. facturer´s instructions. [14] Rohde, M. et al. (2005) Members of the heat-shock protein 70 family promote cancer cell growth by distinct mechanisms. Genes Dev. 19, 570–582, http:// dx.doi.org/10.1101/gad.305405. Conflict of interest [15] Noonan, E., Giardina, C. and Hightower, L. (2008) Hsp70B0 and Hsp72 form a complex in stressed human colon cells and each contributes to cytoprotection. Exp. Cell Res. 314, 2468–2476, http://dx.doi.org/10.1016/j.yexcr.2008.05.002. The authors are employed by Roche Pharma Research and Early [16] Sittler, A. et al. (2001) Geldanamycin activates a heat shock response and Development. Roche has an interest in the development of novel inhibits huntingtin aggregation in a cell culture model of Huntington’s cancer therapies. disease. Hum. Mol. Genet. 10, 1307–1315. [17] Leung, T.K., Hall, C., Rajendran, M., Spurr, N.K. and Lim, L. (1992) The human heat-shock genes HSPA6 and HSPA7 are both expressed and localize to Acknowledgment chromosome 1. Genomics 12, 74–79. [18] Mak, I.W., Liu, L., Ling, V. and Kastelic, T. (2007) The effect of the fungal metabolite radicicol analog A on mRNA degradation. Genomics 90, 723–732, P.K. is supported by the Roche Postdoctoral Fellowship Program http://dx.doi.org/10.1016/j.ygeno.2007.08.002. (RPF). [19] Taiyab, A., Sreedhar, A.S. and Rao, Ch.M. (2009) Hsp90 inhibitors, GA and 17AAG, lead to ER stress-induced apoptosis in rat histiocytoma. Biochem. Pharmacol. 78, 142–152, http://dx.doi.org/10.1016/j.bcp.2009.04.001. Appendix A. Supplementary data [20] Lawson, B., Brewer, J.W. and Hendershot, L.M. (1998) Geldanamycin, an hsp90/GRP94-binding drug, induces increased transcription of endoplasmic Supplementary data associated with this article can be found, in reticulum (ER) chaperones via the ER stress pathway. J. Cell. Physiol. 174, 170– 178, http://dx.doi.org/10.1002/(SICI)1097-4652(199802)174:2<170::AID- the online version, at http://dx.doi.org/10.1016/j.febslet.2015.04. JCP4>3.0.CO;2-L. 053. [21] Li, W.W., Alexandre, S., Cao, X. and Lee, A.S. (1993) Transactivation of the grp78 promoter by Ca2+ depletion. A comparative analysis with A23187 and the endoplasmic reticulum Ca(2+)-ATPase inhibitor thapsigargin. J. Biol. References Chem. 268, 12003–12009. [22] Lytton, J., Westlin, M. and Hanley, M.R. (1991) Thapsigargin inhibits the [1] Stebbins, C.E. et al. (1997) Crystal structure of an Hsp90-geldanamycin sarcoplasmic or endoplasmic reticulum Ca-ATPase family of calcium pumps. J. complex: targeting of a protein chaperone by an antitumor agent. Cell 89, Biol. Chem. 266, 17067–17071. 239–250. [23] Liu, E.S., Ou, J.H. and Lee, A.S. (1992) Brefeldin A as a regulator of grp78 gene [2] Whitesell, L., Mimnaugh, E.G., De Costa, B., Myers, C.E. and Neckers, L.M. expression in mammalian cells. J. Biol. Chem. 267, 7128–7133. (1994) Inhibition of heat shock protein HSP90-pp60v-src heteroprotein [24] Fujiwara, T., Oda, K., Yokota, S., Takatsuki, A. and Ikehara, Y. (1988) Brefeldin A complex formation by benzoquinone ansamycins: essential role for stress causes disassembly of the Golgi complex and accumulation of secretory proteins in oncogenic transformation. Proc. Natl. Acad. Sci. USA 91, 8324– proteins in the endoplasmic reticulum. J. Biol. Chem. 263, 18545–18552. 8328. [25] Ramirez, V.P., Stamatis, M., Shmukler, A. and Aneskievich, B.J. (2014) Basal and [3] Trepel, J., Mollapour, M., Giaccone, G. and Neckers, L. (2010) Targeting the stress-inducible expression of HSPA6 in human keratinocytes is regulated by dynamic HSP90 complex in cancer. Nat. Rev. Cancer 10, 537–549, http:// negative and positive promoter regions. Cell Stress Chaperones, http:// dx.doi.org/10.1038/nrc2887. dx.doi.org/10.1007/s12192-014-0529-0. [4] Hong, D.S. et al. (2013) Targeting the molecular chaperone heat shock protein [26] Gano, J.J. and Simon, J.A. (2010) A proteomic investigation of ligand- 90 (HSP90): lessons learned and future directions. Cancer Treat. Rev. 39, 375– dependent HSP90 complexes reveals CHORDC1 as a novel ADP-dependent 387, http://dx.doi.org/10.1016/j.ctrv.2012.10.001. HSP90-interacting protein. Mol. Cell. Proteomics 9, 255–270, http:// [5] Garcia-Carbonero, R., Carnero, A. and Paz-Ares, L. (2013) Inhibition of HSP90 dx.doi.org/10.1074/mcp.M900261-MCP200. molecular chaperones: moving into the clinic. Lancet Oncol. 14, e358–369, [27] Noonan, E.J., Place, R.F., Giardina, C. and Hightower, L.E. (2007) Hsp70B0 http://dx.doi.org/10.1016/S1470-2045(13)70169-4. regulation and function. Cell Stress Chaperones 12, 393–402. [6] Jhaveri, K., Taldone, T., Modi, S. and Chiosis, G. (1823) Advances in the clinical [28] Noonan, E.J., Fournier, G. and Hightower, L.E. (2008) Surface expression of development of heat shock protein 90 (Hsp90) inhibitors in cancers. Biochim. Hsp70B0 in response to proteasome inhibition in human colon cells. Cell Stress Biophys. Acta 742–755, 2012, http://dx.doi.org/10.1016/ Chaperones 13, 105–110, http://dx.doi.org/10.1007/s12192-007-0003-3. j.bbamcr.2011.10.008. [29] Smith, K.J. et al. (2010) Heat shock protein 70B0 (HSP70B0) expression and [7] Modi, S. et al. (2011) HSP90 inhibition is effective in breast cancer: a phase II release in response to human oxidized low density lipoprotein immune trial of tanespimycin (17-AAG) plus trastuzumab in patients with HER2- complexes in macrophages. J. Biol. Chem. 285, 15985–15993, http:// positive metastatic breast cancer progressing on trastuzumab. Clin. Cancer dx.doi.org/10.1074/jbc.M110.113605. Res. 17, 5132–5139, http://dx.doi.org/10.1158/1078-0432.CCR-11-0072. [30] Khalouei, S., Chow, A.M. and Brown, I.R. (2014) Stress-induced localization of [8] Xu, W. et al. (2001) Sensitivity of mature Erbb2 to geldanamycin is conferred HSPA6 (HSP70B0) and HSPA1A (HSP70-1) proteins to centrioles in human by its kinase domain and is mediated by the chaperone protein Hsp90. J. Biol. neuronal cells. Cell Stress Chaperones 19, 321–327, http://dx.doi.org/10.1007/ Chem. 276, 3702–3708, http://dx.doi.org/10.1074/jbc.M006864200. s12192-013-0459-2. [9] Bagatell, R. et al. (2000) Induction of a heat shock factor 1-dependent stress [31] Khalouei, S., Chow, A.M. and Brown, I.R. (2014) Localization of heat shock response alters the cytotoxic activity of hsp90-binding agents. Clin. Cancer protein HSPA6 (HSP70B0) to sites of transcription in cultured differentiated Res. 6, 3312–3318. human neuronal cells following thermal stress. J. Neurochem. 131, 743–754, [10] Ma, L. et al. (2014) Dual targeting of heat shock proteins 90 and 70 promotes http://dx.doi.org/10.1111/jnc.12970. cell death and enhances the anticancer effect of chemotherapeutic agents in [32] Schneider, B. et al. (2012) Targeted siRNA delivery and mRNA knockdown bladder cancer. Oncol. Rep. 31, 2482–2492, http://dx.doi.org/10.3892/ mediated by bispecific digoxigenin-binding antibodies. Mol. Ther. Nucleic or.2014.3132. Acids 1, e46, http://dx.doi.org/10.1038/mtna.2012.39. [11] Hageman, J., van Waarde, M.A., Zylicz, A., Walerych, D. and Kampinga, H.H. (2011) The diverse members of the mammalian HSP70 machine show distinct